Abstract
This paper describes a study on the relationship of the convergence perfoemance between the error signal and the modeling error of the plant (shaking table for earthquake simulators) with iterative learning control (ILC) system. The error signal converge to zero with minimum number of trials by using inverse model of the plant as a learning law. However, due to the existance of the modeling error caused by a specimen on the shaking table, it is difficult to achieve the best performance of ILC. So, we verified the relationship between the modeling error and the convergence performance of ILC by using numerical simulations and experiments with prototype system.
